With popularization of a telecommunication cloud concept, a layered architecture in which a service application layer is separated or decouple from an IaaS (Infrastructure as a Service) layer is widely applied to a telecommunication system. In the telecommunication system, hardware devices are managed by the IaaS layer, which are transparent to service applications, and the service applications generally operate in VMs (Virtual Machine), as shown in FIG. 1.
A typical characteristic of telecommunication applications is high reliability, and the high reliability is achieved in the telecommunication system by means of cooperation between a master process and a slave process. As shown in FIG. 2A, a master service process is run on VM1, and handles services; a slave service process is run on VM2, but does not handle any service. Status and data of the slave service process are synchronized from the master service process. Processes running in the system are monitored by a management process deployed on VM3, if the management process monitors that an exception occurs on a service process, then it will notify the slave service process on VM2 to convert to a master service process for handling services, whereas the master service process on VM1 converts to a slave service process, thereby achieving service restoration, as shown in FIG. 2B.
Under the layered architecture, in order to guarantee reliability of telecommunication applications over the cloud, if resource status in the IaaS layer changes, for instance, hardware is failure or a Host OS (Host Operating System, an operating system operating in a host) running VMs crashes, a related notification capability needs to be provided in the IaaS layer so as to handle an exception event, and operations such as service migration and failure isolation are completed by the IaaS layer together to provide telecommunication services of high reliability to a user.
Generally, when notify or handle an exception event notification, as shown in FIG. 3, when an IaaS agent process detects that a VM fails, the IaaS reports the failure to an IaaS management central node, the IaaS management central node then notifies an application layer management central node, and finally the application layer management central node notifies an application layer management process.
For the above process for notifying and handling the exception event notification, the channel path is relatively long, the exception event cannot be notified to the application layer in time, and the application layer management central node and the IaaS management central node, both of which are management and maintenance nodes, serve as failure notification channels, thus the reliability is relatively low.